Fin cleaning material conveying system and task allocation method

Abstract

The invention relates to the technical field of conveying and task allocation, and provides a fin cleaning material conveying system and task allocation method.Through organic combination of a modal sensing module, a task allocation module, a path control module and a feedback monitoring module, the highly-intelligent and collaborative conveying system is constructed, information interaction among all the modules is smooth, and the efficiency is improved. Conveying and task distribution can be adjusted in real time according to the change of the working environment and the state of the robot, and the overall efficiency and quality of fin cleaning material conveying and task distribution are greatly improved.

Description

Technical Field

[0001] The present application relates to the technical field of conveying and task allocation, and in particular to a fin cleaning material conveying system and a task allocation method. Background Art

[0002] In industrial production, fin cleaning and material transportation are important links to ensure the efficient operation of production equipment. Traditional systems often rely on only a single type of sensor and cannot fully obtain multi-dimensional information about the area environment that needs to be cleaned. This makes it difficult to accurately identify the location of obstacles, accurately assess the degree of fin contamination, and quantify the difficulty of cleaning. At the same time, there is a lack of comprehensive consideration of the robot's own status, and often only focuses on the robot's movement ability, ignoring factors such as joint availability and regional accessibility. This makes task allocation unreasonable, and it is easy for the robot to be unable to complete complex cleaning tasks due to joint failure. The traditional task allocation method does not fully take into account the actual situation of the work area, resulting in low cleaning efficiency and chaotic task execution.

[0003] Similarly, existing systems typically use fixed path planning algorithms that fail to dynamically adjust based on real-time obstacle locations and the complexity of fin structures. This makes robots prone to collisions in complex working environments, and when faced with fins of varying structures, they are unable to complete cleaning at the optimal path and speed, impacting cleaning effectiveness and efficiency. Furthermore, these systems fail to provide timely and effective comprehensive monitoring of the robot's working status during task execution. When abnormal working conditions occur, feedback and parameter updates are not promptly provided, leading to persistent problems and even cascading failures, impacting the stability of the entire material handling and cleaning process. Consequently, they fail to address the problem of how to implement task allocation, path planning, and feedback adjustments through environmental perception and equipment status analysis. Summary of the Invention

[0004] In response to the deficiencies of the prior art, the present application provides a fin cleaning material conveying system and a task allocation method.

[0005] In a first aspect, the present application provides a fin cleaning material conveying system, the system comprising: a modal perception module, a task allocation module, a path control module, and a feedback monitoring module;

[0006] The modal perception module is used to obtain the area position, contaminant thickness and fin surface pressure through the sensor array, and fuse the area position, contaminant thickness and fin surface pressure to construct a contamination area map. The contamination area map includes the obstacle position, the size of the working area, the degree of fin contamination and the difficulty of fin cleaning. The contaminant thickness is obtained by visual detection.

[0007] The task assignment module is used to obtain the remaining power, joint availability, and area accessibility of each robot in real time to construct a state vector for each robot, divide the tasks according to the location of the area in the defaced area map, determine the matching degree between each robot and the task, generate task instructions for the robot based on the matching degree, and coordinate the working order of the robots;

[0008] The path control module is used to extract the task instructions of each robot, determine the task position of each robot, and dynamically plan the path and speed of each robot based on the obstacle position of the contaminated area map to control each robot to perform the task;

[0009] The feedback monitoring module is used to monitor the working status of each robot in performing tasks, and when an abnormal working status occurs, feedback is provided to update the contaminated area map and the state vector of each robot.

[0010] As an optional implementation, the construction logic of the contaminated area map includes:

[0011] The zone location, contaminant thickness, and fin surface pressure obtained by the sensor array are aligned in time and space, and the obstacle location is identified at the zone location to mark the inaccessible area and the working area, and the size of the working area is determined;

[0012] In the working area, the degree of fin contamination is comprehensively judged based on the thickness of the contaminants and the fin surface pressure, and the complexity of the fin structure in the working area is extracted;

[0013] The difficulty of fin cleaning is determined by the degree of fin contamination and the complexity of the fin structure.

[0014] As an optional implementation, the logic for constructing the state vector of each robot includes:

[0015] By monitoring the power of each robot, the remaining power of each robot can be obtained;

[0016] Monitor the joint angle deviation of each robot to obtain the joint availability of each robot;

[0017] The remaining power, joint availability, and area reachability of each robot are weightedly fused to obtain the state vector of each robot.

[0018] As an optional implementation, the logic for obtaining the area reachability of each robot includes:

[0019] Combined with the obstacle locations in the contaminated area map, a safe path is searched for each robot within the working area.

[0020] Divide the safe path into multiple points and calculate the minimum distance from each point on the safe path to the obstacle location;

[0021] The safety threshold of each robot is set according to its structure, and the accessibility coefficient of each robot in the working area is determined according to the minimum distance from each point to the obstacle position and the safety threshold of each robot;

[0022] The fitness factor of each robot is determined according to its structure, and the regional reachability of each robot is obtained by comprehensive judgment based on the reachability coefficient of each robot in the working area and the fitness factor of each robot.

[0023] As an optional implementation, the logic for determining the matching degree between each robot and the task includes:

[0024] Divide the contaminated area map into tasks according to the area location and construct a task requirement vector, which includes the task cleaning difficulty and task urgency;

[0025] Determine the capability matching based on the task requirement vector and the state vector of each robot;

[0026] Determine the path length of each robot to the task area, and combine the fin structure complexity to obtain the motion path cost of each robot;

[0027] The matching degree of each robot to the task is judged based on the matching degree of capabilities and the motion path cost of each robot.

[0028] As an optional implementation, the coordination logic of the robot's working sequence includes:

[0029] Arrange tasks in descending order of urgency to generate a time window sequence;

[0030] Determine the temporal continuity of each robot as it completes the time window sequence;

[0031] The tasks that each robot completes in the time window sequence are determined based on the matching degree and time continuity between each robot and the task, so as to coordinate the working order of the robots.

[0032] As an optional implementation, the planning logic for the path and speed of each robot includes:

[0033] Extract the task instructions of each robot to determine the task position and starting position of each robot, and combine the boundary information of the working area in the defaced area map to generate the initial path and initial speed of each robot;

[0034] Adjust the dwell time and coverage of the initial path according to the difficulty of fin cleaning;

[0035] The initial path and initial speed are optimized according to the complexity of the fin structure to obtain the path and speed of each robot;

[0036] The position information of each robot is obtained in real time, and the path and speed of each robot are shared. It is determined whether there is a path conflict to update the path and speed of each robot.

[0037] As an optional implementation, the update logic of the contaminated area map includes:

[0038] Monitor whether there are new obstacles to update the obstacle location;

[0039] After each robot performs a task, the contaminant thickness and fin surface pressure in the task area are monitored;

[0040] According to the change of the thickness of the pollutants in the task area, it is judged whether the task area is an abnormal working area, and the degree of fin contamination is adjusted in combination with the change of the fin surface pressure in the task area;

[0041] If the task area is an abnormal working area twice in a row, the difficulty of fin cleaning will increase.

[0042] As an optional implementation, the distance from each point on the safe path to all obstacle locations in the contaminated area map is calculated using the Euclidean distance formula, and the minimum value of the distance is selected as the minimum distance from each point to the obstacle location.

[0043] As an optional implementation, the sub-logic for obtaining the joint availability of each robot includes:

[0044] Monitor the actual angle of each joint in real time, calculate the absolute value of the difference between the actual angle of each joint and the preset nominal angle, and obtain the joint angle deviation of each joint;

[0045] Calculate the ratio of the joint angle deviation of each joint to the maximum angle deviation of each joint to obtain the joint deviation of each joint;

[0046] Subtract the joint deviation of each joint from 1 to obtain the joint availability of each joint;

[0047] The joint availability of all joints of each robot is averaged to obtain the joint availability of each robot.

[0048] In a second aspect, the present application provides a method for allocating material delivery tasks for fin cleaning, the method comprising: acquiring a region position, a contaminant thickness, and a fin surface pressure through a sensor array, fusing the region position, the contaminant thickness, and the fin surface pressure to construct a contamination region map, the contamination region map including an obstacle location, a size of a working area, a degree of fin contamination, and a difficulty level for fin cleaning;

[0049] Obtain the remaining power, joint availability, and area reachability of each robot in real time to construct the state vector of each robot;

[0050] Divide the defaced area map into tasks according to the area location, determine the matching degree between each robot and the task, generate task instructions for the robot based on the matching degree, and coordinate the working order of the robots;

[0051] Extract the task instructions of each robot, determine the task location of each robot, and dynamically plan the path and speed of each robot based on the obstacle location of the contaminated area map to control each robot to perform the task;

[0052] Monitor the working status of each robot in performing tasks. When an abnormal working status occurs, feedback is provided and the contaminated area map and the state vector of each robot are updated.

[0053] Compared with the existing technology, the beneficial effect of this application is: through the organic combination of modal perception module, task allocation module, path control module and feedback monitoring module, a highly intelligent and collaborative conveying system is constructed, and information between modules interacts smoothly, and the conveying and task allocation can be adjusted in real time according to changes in the working environment and the status of the robot, greatly improving the overall efficiency and quality of fin cleaning material conveying and task allocation.

[0054] The modal perception module acquires multi-dimensional information through the sensor array and integrates it to construct a map of the contaminated area, comprehensively and accurately presenting the various characteristics of the working area. It can accurately identify the location of obstacles, clearly divide the size of the working area, accurately judge the degree of fin contamination and quantify the difficulty of fin cleaning, providing a rich and reliable data foundation for subsequent modules, enabling the entire system to have a deeper and more accurate understanding of the working environment.

[0055] The task allocation module obtains the robot's status information in real time to construct a state vector, and combines it with the contaminated area map to perform scientific task division and matching judgment. It can reasonably allocate tasks to achieve the best match between robots and tasks, improve the robot's work efficiency and task completion quality, and at the same time coordinate the robot's work order to avoid task conflicts and confusion, and optimize the multi-robot collaborative operation process.

[0056] The path control module dynamically plans the robot's path and speed based on task instructions and the obstacle locations in the contaminated area map, enabling the robot to perform tasks safely and efficiently in complex environments, avoid collisions, and improve cleaning effects. It adjusts the path and speed according to the difficulty of fin cleaning and the complexity of the fin structure, further optimizing the cleaning process and improving cleaning quality and efficiency.

[0057] The feedback monitoring module monitors the working status of the robot in real time. When an abnormality occurs, it quickly feeds back and updates the contaminated area map and robot state vector. It can promptly discover and solve problems, ensure the stable operation of the system, enable the system to adjust its strategy in time according to the latest situation, and improve the reliability and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be derived from these drawings without inventive work. Among them:

[0059] Figure 1 This is a system flow chart of the fin cleaning material conveying system provided in an embodiment of the present application;

[0060] Figure 2 Constructing a logic diagram for the contaminated area map of the fin cleaning material delivery system provided in the embodiment of the present application;

[0061] Figure 3 A logic diagram for judging the matching degree between each robot and the task of the fin cleaning material conveying system provided in an embodiment of the present application;

[0062] Figure 4 A working sequence coordination logic diagram of a robot in a fin cleaning material conveying system provided in an embodiment of the present application;

[0063] Figure 5 This is a flow chart of the method for allocating fin cleaning material delivery tasks provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the objectives, technical solutions and advantages of the embodiments of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application are clearly and completely described below in conjunction with the drawings in the specification. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0065] Example 1:

[0066] like Figure 1 As shown, a system flow chart of a fin cleaning material conveying system is provided for an embodiment of the present application. The system includes a modal perception module, a task allocation module, a path control module and a feedback monitoring module.

[0067] The modal perception module is used to obtain the area position, contaminant thickness and fin surface pressure through the sensor array, and fuse the area position, contaminant thickness and fin surface pressure to construct a contaminated area map. The contaminated area map includes the obstacle location, the size of the working area, the degree of fin contamination and the difficulty of fin cleaning. The contaminant thickness is obtained through visual inspection.

[0068] Specifically, if Figure 2 As shown in the figure, the construction logic of the contaminated area map includes:

[0069] The zone location, contaminant thickness, and fin surface pressure obtained by the sensor array are aligned in time and space, and the obstacle location is identified at the zone location to mark the inaccessible area and the working area, and the size of the working area is determined;

[0070] In the working area, the degree of fin contamination is comprehensively judged based on the thickness of the contaminants and the fin surface pressure, and the complexity of the fin structure in the working area is extracted;

[0071] The difficulty of fin cleaning is determined by the degree of fin contamination and the complexity of the fin structure.

[0072] The data obtained by different types of sensors differ in time and space. If they are not aligned, the data cannot be accurately fused to construct a map of the contaminated area, which will lead to errors in the association of the area position, contaminant thickness and fin surface pressure. The laser radar emits a laser beam and receives the reflected light, and obtains the distance information of the surrounding environment by measuring the flight time of the light, generating a large amount of three-dimensional point cloud data to determine the area position. For the contaminant thickness detected by the infrared thermal imager and the fin surface pressure measured by the flexible tactile sensor, a timestamp-based synchronization algorithm is used to add an accurate timestamp to each data point when the sensor obtains the data. Through timestamp matching, the data obtained by different sensors at the same time are associated, and the interpolation algorithm is used to preprocess the data to make the data from different sensors consistent in time and space resolution. Through time and space alignment, the area position, contaminant thickness and fin surface pressure from different sensors can be accurately associated, providing an accurate data basis for the subsequent construction of the contaminated area map.

[0073] Obstacle locations are identified through 3D semantic segmentation at the regional locations to mark the inaccessible area and the working area. The inaccessible area is the area with obstacles, and the working area is the area where work can be done. The size of the working area is also determined simultaneously. At the same time, the data from a single sensor cannot fully and accurately reflect the degree of fin contamination. Therefore, within the working area, the degree of fin contamination is comprehensively judged based on the contaminant thickness and the fin surface pressure. The functional expression of the fin contamination degree is as follows:

[0074] D(p)=k1×L(p)+k2×F(p)+k3×ΔT(p);

[0075] Where D(p) represents the degree of fin contamination, L(p) represents the thickness of the contaminant, F(p) represents the fin surface pressure, ΔT(p) represents the difference between the current temperature and the target temperature, k1, k2, and k3 represent the weight values of the contaminant thickness, the fin surface pressure, and the difference between the current temperature and the target temperature, respectively.

[0076] It should be noted that: in the early stage of system operation, the initial weight values of k1, k2 and k3 are set based on experience, and then the weight values are continuously updated through a large amount of experimental data and actual cleaning cases through the Bayesian inference algorithm. For example, when F(p) / L(p)>θ, it is judged as viscous pollutants, and the weight value of k2 is increased to highlight the influence of pressure on the degree of fin fouling, where θ is the threshold set through experience; and when ΔT(p) exceeds the difference threshold, it is judged as warm coking type fouling, and the weight value of k3 is increased to strengthen the role of temperature factors in the calculation of the degree of fin fouling.

[0077] Through multimodal data fusion and dynamic weight allocation, the degree of fin contamination can be quantified more comprehensively and accurately, fully considering the impact of different pollutant characteristics and fin surface conditions on the degree of fin contamination. Compared with the traditional fixed weight fusion method, it can more flexibly adapt to different contamination conditions and improve the accuracy of quantification of the degree of fin contamination. Through three-dimensional semantic segmentation, different areas in the environment can be identified, providing important environmental information for the robot's path planning and task allocation.

[0078] Then, the complexity of the fin structure is extracted through image processing and pattern recognition technology. The fin surface image is obtained through an industrial camera and grayscale processing is performed. The noise interference of the fin surface image is removed by the Gaussian filtering algorithm. The fin surface image is smoothed while retaining the structural edge information of the fin. Then, the canny edge detection algorithm is used to calculate the gradient amplitude and gradient direction to accurately detect the contour edge of the fin. The detected edge is morphologically processed, and the dilation and erosion operations are used to repair the broken parts of the edge and remove isolated noise points. The fins include corrugated fins and windowed fins.

[0079] For corrugated fins, by analyzing the contour edge of the fin, the edge information of the spatial domain is converted to the frequency domain using Fourier transform. In the frequency domain, the peak frequency corresponding to the corrugation period is found to calculate the corrugation period. For example, discrete Fourier transform is performed on the edge curve, and the frequency corresponding to the main peak is determined by analyzing the spectrum diagram. Then, the corrugation period is calculated based on the relationship between frequency and period. For windowed fins, morphological operations and connected domain analysis are used to identify the window area, calculate the number and area of the windows, and then obtain the window density. For example, a binary image is used to separate the window area from the fin matrix, and then a connected domain labeling algorithm is used to statistically analyze the window area. The number of windows is obtained by calculating the ratio of the number of pixels in the window area to the number of pixels in the entire fin image to obtain the window density; the surface roughness characteristics of the fin are extracted, and the standard deviation of the grayscale values is calculated as an approximate measure of surface roughness through statistical analysis of the grayscale values of the fin surface. For example, on the preprocessed grayscale image, a local area of a certain size is selected, and the standard deviation of the grayscale values in the area is calculated. The overall surface roughness of the fin is obtained through statistical analysis of multiple local areas; thus, the key parameters reflecting the complexity of the fin structure, such as the corrugation period, window density and surface roughness, are accurately calculated, providing a specific data basis for quantifying the complexity of the fin structure.

[0080] The calculated structural characteristic parameters are integrated to form a comprehensive quantitative index of fin structure complexity, which is convenient for unified analysis and application in the entire system. For corrugated fins, the fin structure complexity = 1 / corrugation period + roughness factor, where the roughness factor is adjusted according to the difference between the surface roughness and the standard roughness. For example, when the surface roughness is greater than the standard roughness by a certain proportion, the roughness factor increases to highlight the impact of surface roughness on structural complexity.

[0081] For window-shaped fins, the fin structure complexity = window density + window shape complexity factor, where the window shape complexity factor is adjusted according to the shape regularity of the window. For example, for irregularly shaped windows, the window shape complexity factor is increased to reflect the contribution of shape complexity to structural complexity. Through experiments and actual application data, the value range and adjustment rules of the roughness factor and the window shape complexity factor are determined to make the quantification results of the fin structure complexity more in line with the actual cleaning difficulty and the complexity of robot operation. In this way, a quantitative indicator that can comprehensively reflect the complexity of the fin structure is obtained, providing unified and accurate structural complexity information for subsequent cleaning difficulty calculations and system decision-making.

[0082] In order to reasonably arrange the robot's cleaning tasks and plan the cleaning path, it is necessary to accurately quantify the fin cleaning difficulty according to the degree of fin contamination and the complexity of the fin structure, so as to assign the task to the most suitable robot and formulate an efficient cleaning strategy; the local cleaning difficulty is obtained by multiplying the degree of fin contamination and the complexity of the fin structure, and the local cleaning difficulty is subjected to regional clustering analysis through the density clustering algorithm. The algorithm can automatically identify different clusters according to the density distribution of data points, and divide the fin cleaning difficulty into three levels: high, medium and low. Then, a detailed cleaning difficulty heat map is generated based on the clustering results to intuitively display the distribution of fin cleaning difficulty in different areas; by comprehensively considering the degree of fin contamination and the complexity of the fin structure, the fin cleaning difficulty can be accurately quantified, which provides a scientific basis for task allocation and path planning. The generated cleaning difficulty heat map intuitively shows the fin cleaning difficulty in different areas, which is convenient for operators and systems to make task decisions and allocate resources.

[0083] During the cleaning process, the regional environment and the status of the fins will constantly change, and the contaminated area map needs to be updated in real time to ensure that the system can adjust task allocation and path planning based on the latest information to improve cleaning efficiency and effectiveness. The system also receives real-time feedback on abnormal working status transmitted by the monitoring module to update the contaminated area map. When a new obstacle is detected, the obstacle position is updated, and then the semantic segmentation results of the regional position, the degree of fin contamination, and the difficulty of fin cleaning are synchronously corrected to ensure that the contaminated area map is always consistent with the actual environment. Through continuous feedback and online updates of the map, changes in the environment and fin status during the cleaning process can be reflected in real time, ensuring that the system always makes decisions based on the latest information. The real-time updated contaminated area map provides the task allocation module and the path control module with the latest environmental and task information. The task allocation module can re-evaluate the matching degree between the task and the robot based on the updated contaminated area map and adjust the task allocation. The path control module can dynamically plan the robot's path and speed based on the latest obstacle location and fin cleaning difficulty, improving cleaning efficiency and effectiveness and ensuring the efficient operation of the entire fin cleaning material conveying system.

[0084] The task allocation module is used to obtain the remaining power, joint availability and area accessibility of each robot in real time to construct the state vector of each robot, divide the tasks according to the area location of the contaminated area map, determine the matching degree between each robot and the task, generate the robot's task instructions according to the matching degree, and coordinate the robot's working order.

[0085] Specifically, the logic for obtaining the area reachability of each robot includes:

[0086] Combined with the obstacle locations in the contaminated area map, a safe path is searched for each robot within the working area.

[0087] Divide the safe path into multiple points and calculate the minimum distance from each point on the safe path to the obstacle location;

[0088] The safety threshold of each robot is set according to its structure, and the accessibility coefficient of each robot in the working area is determined according to the minimum distance from each point to the obstacle position and the safety threshold of each robot;

[0089] The fitness factor of each robot is determined according to the type of each robot, and the regional reachability of each robot is obtained by comprehensive judgment based on the accessibility coefficient of each robot in the working area and the fitness factor of each robot.

[0090] In order to ensure that the robot can move safely and complete tasks in complex working areas, it is necessary to find a feasible path that avoids obstacles. This is the basis for evaluating whether the robot can reach the task area and perform the task. Through the improved A* search algorithm, the obstacle position in the contaminated area map is used as the search constraint. During the search process, the A* search algorithm considers the robot's kinematic characteristics, such as turning radius and minimum moving distance. For example, for wheeled robots, the path curvature is ensured to meet its turning radius limit during path search. At the same time, in order to improve the search efficiency, a two-way search strategy is adopted, searching from the starting point and the target point at the same time. When the nodes of the two search directions meet, a safe path is found. In this way, a safe path can be searched for each robot in a complex working area quickly and accurately, obstacles can be effectively avoided, and the safety of the robot in the process of performing the task can be ensured.

[0091] By calculating the minimum distance from each point on the safe path to the obstacle location, the robot's safety on the path is quantified, providing data support for the subsequent determination of the accessibility coefficient, so as to more accurately evaluate the robot's accessibility in the working area; the safe path is discretized into a series of dense points, and for each point, the distance from its to all obstacle locations in the contaminated area map is calculated using the Euclidean distance formula, and then the minimum value is selected from all calculated distances as the minimum distance from the point to the obstacle location; parallel computing technology is adopted, and the robot's multi-core processor is utilized to simultaneously calculate the minimum distance of multiple points on the path, greatly improving the computing efficiency. For large-scale path point sets, this parallel computing method can significantly shorten the calculation time; thereby accurately calculating the minimum distance from each point on the safe path to the obstacle location, and comprehensively quantifying the safety of the path. The application of parallel computing technology enables fast completion of distance calculations even in the case of complex paths and a large number of obstacles, improving the real-time response capability of the system.

[0092] The minimum distance is compared with the safety threshold and converted into a quantitative accessibility coefficient to intuitively measure the robot's accessibility within the work area, providing a standardized indicator for comprehensive assessment of regional accessibility. A corresponding safety threshold is set according to the structure of each robot. For example, for small and lightweight robots, the safety threshold is set relatively high due to their weak collision resistance, while for large and sturdy robots, the safety threshold can be appropriately reduced. The specific value of the safety threshold is determined through experimental testing and theoretical analysis. For each point on the safe path, the minimum distance is divided by the safety threshold of the corresponding robot to obtain the unreachable value. The unreachable value is subtracted from 1 to obtain the accessibility coefficient of the corresponding robot. The accessibility coefficients of all points on the safe path are statistically analyzed, such as calculating the mean and median, to obtain the overall accessibility coefficient of the robot on this safe path. The accessibility coefficient clearly quantifies the robot's accessibility along a specific safe path within the work area, making the accessibility of different robots on different paths comparable, and providing an intuitive and standardized indicator for subsequent comprehensive assessment of regional accessibility.

[0093] Different types of robots have different adaptability under different environments and task conditions. By determining the fitness factor and combining it with the accessibility coefficient, the actual accessibility of the robot in the working area can be more comprehensively evaluated, providing a more accurate basis for task allocation. The fitness factor is determined according to the type of robot. For example, for a wheeled robot, the fitness factor is calculated based on the flatness and slope of the ground. The fitness factor of the wheeled robot can be calculated by 1-(ground flatness deviation + absolute value of slope) / (maximum allowable flatness deviation + maximum allowable slope). For a wall-climbing robot, the fitness factor is calculated based on the relationship between its adsorption force and its own weight and the characteristics of the wall material. For example, magnetic adsorption force × wall friction coefficient / self-weight can be used to obtain the fitness factor of the wall-climbing robot, where the wall friction coefficient is The coefficient is obtained by experimentally measuring the friction coefficient between different wall materials and the robot adsorption device; then the regional accessibility of each robot is obtained by multiplying the accessibility coefficient of each robot in the working area by the fitness factor of each robot; thus, by comprehensively considering the accessibility coefficient and the fitness factor, the actual accessibility of the robot in the working area is comprehensively evaluated, and the physical characteristics of the robot and the task environment factors are fully considered, so that the evaluation results are more in line with the actual situation, and a more accurate accessibility basis is provided for task allocation. Accurate regional accessibility is used to construct the robot's state vector, so that the state vector can more realistically reflect the robot's actual working ability and state, and provide important state information for the subsequent judgment of the matching degree between the robot and the task, which plays a key role in the reasonable allocation of tasks.

[0094] Specifically, the construction logic of each robot's state vector includes:

[0095] By monitoring the power of each robot, the remaining power of each robot can be obtained;

[0096] Monitor the joint angle deviation of each robot to obtain the joint availability of each robot;

[0097] The remaining power, joint availability, and area reachability of each robot are weightedly fused to obtain the state vector of each robot.

[0098] The remaining power is an important indicator for measuring the robot's working endurance, which directly affects the time and range of the robot's ability to perform tasks. In the robot's power management system, the voltage and current parameters of the power battery are monitored in real time to obtain the current power, basic power consumption and maximum power of each robot. The remaining power of each robot is obtained by dividing the difference between the current power and basic power consumption by the difference between the maximum power and basic power consumption. This allows the remaining power of each robot to be accurately obtained, providing reliable data for evaluating the robot's working endurance, enabling the system to promptly understand the robot's energy status, reasonably arrange tasks, and avoid task interruptions due to insufficient power.

[0099] The health of a robot's joints directly affects its motion capabilities and the accuracy of its task execution. Monitoring joint angle deviation to obtain joint availability can assess the robot's motion function status and is an indispensable part of constructing the state vector, used to determine the robot's ability to perform complex tasks. An angle sensor is installed at each robot joint to monitor the actual joint angle in real time. The actual angle is subtracted from the preset nominal angle to obtain the joint angle deviation. For each joint, the joint availability is calculated based on its maximum allowable angle deviation. The formula for calculating joint availability is as follows:

[0100] D m =1-|λ m -λ0| / λ max ;

[0101] Where D m represents the joint availability of the m-th joint, λ m represents the actual angle of the mth joint, λ0 represents the preset nominal angle, and λ max Indicates the maximum angular deviation.

[0102] This enables accurate monitoring of the health status of each robot joint. The joint availability of each robot is the average value of the joint availability of all the robot's joints. By quantitatively evaluating the robot's motion function status through joint availability, potential joint problems can be discovered in a timely manner, providing support for ensuring that the robot can perform its tasks normally.

[0103] The remaining power, joint availability, and regional accessibility are weighted and fused to generate a comprehensive state vector that comprehensively reflects the robot's working ability and status, providing a unified quantitative indicator for judging the match between the robot and the task, making task allocation more reasonable; the remaining power, joint availability, and regional accessibility of each robot are weighted and fused to obtain the state vector of each robot, where the weight coefficients of the remaining power, joint availability, and regional accessibility are dynamically adjusted through the reinforcement learning algorithm, with the robot's success rate in completing the task, task execution time, and power consumption as the reward function. For example, when the robot successfully completes the task with low power consumption and short execution time, a higher reward is given, and the weight coefficient is continuously adjusted based on the reward feedback; through dynamic weighted fusion, the generated state vector can more flexibly and accurately reflect the actual working status of the robot, adapt to different tasks and environmental requirements, and the application of the reinforcement learning algorithm enables the weight coefficient to be continuously optimized according to the actual performance of the robot, improving the effectiveness and rationality of the state vector and providing a more accurate basis for task allocation.

[0104] Specifically, if Figure 3 As shown, the judgment logic for the matching degree between each robot and the task includes:

[0105] Divide the contaminated area map into tasks according to the area location and construct a task requirement vector, which includes the task cleaning difficulty and task urgency;

[0106] Determine the capability matching based on the task requirement vector and the state vector of each robot;

[0107] Determine the path length of each robot to the task area, and combine the fin structure complexity to obtain the motion path cost of each robot;

[0108] The matching degree of each robot to the task is judged based on the matching degree of capabilities and the motion path cost of each robot.

[0109] In order to accurately describe the characteristics and requirements of each task so as to match it with the robot's state vector, it is necessary to construct a task requirement vector containing key task information as an important basis for judging the matching degree between the robot and the task; the contaminated area map is divided into several tasks according to the regional location, and the relevant information of the task area is extracted from the contaminated area map to construct a task requirement vector. The task requirement vector includes the task cleaning difficulty and task urgency, where the task cleaning difficulty is obtained by judging the fin cleaning difficulty in the task area; the task urgency is obtained according to the inverse of the pollutant diffusion rate, where the pollutant diffusion rate is monitored in real time by sensors; the task requirement vector thus constructed comprehensively and accurately describes the key characteristics and requirements of the task, and through the task cleaning difficulty and task urgency, provides clear and definite task information for subsequent matching with the robot's state vector, making the matching judgment more scientific.

[0110] By comparing the task requirement vector and the robot's state vector, the degree of match between the robot's capabilities and the task is determined, and whether the robot has the ability to complete the task is evaluated, providing an important basis for judging the match between the robot and the task; based on the task requirement vector and the robot's state vector, the vector inner product method is used to calculate the capability match. In this way, the robot's remaining power, joint availability, and area accessibility are comprehensively compared with the task cleaning difficulty and task urgency; to improve the accuracy of the capability match calculation, the task requirement vector and the robot's state vector are normalized, and the values of each element in the task requirement vector and the robot's state vector are mapped to the interval [0,1] to eliminate the impact of differences in parameter value ranges on the calculation results; thus, the degree of match between the robot's capabilities and the task can be quantified. Through normalization and vector inner product calculation, the robot state and task requirements are comprehensively and objectively compared, providing an important capability match basis for judging the match between the robot and the task, and improving the accuracy of the match judgment.

[0111] Taking into account the path length from the robot to the task area and the complexity of the fin structure, the motion path cost of the robot to perform the task is determined, and the cost of the robot to perform the task under spatial and environmental factors is evaluated, providing more comprehensive information for comprehensively judging the matching degree between the robot and the task; the path length of each robot to the location of the task area is determined by a path planning algorithm (such as the Dijkstra algorithm), and when calculating the path length, the robot's kinematic constraints and obstacles in the environment are considered. For example, for a wheeled robot, when searching for the path, the curvature of the path is ensured to meet its turning radius limit, and the motion path cost is calculated in combination with the complexity of the fin structure to obtain the path cost = path length × fin structure complexity; thereby accurately quantifying the motion path cost of the robot to perform the task, comprehensively considering the path length and the complexity of the fin structure, providing a quantitative indicator for evaluating the cost of the robot to perform the task in the actual working environment, making the judgment of the matching degree between the robot and the task more comprehensive and accurate.

[0112] The ability matching degree and motion path cost are comprehensively considered to comprehensively judge the matching degree between the robot and the task, providing the final decision basis for task allocation, ensuring that the task is assigned to the most suitable robot, and improving the efficiency and quality of task execution. The function expression defining the matching degree is as follows:

[0113] M ij =σ(α×(R i ·T j )-β×(W ij ×P));

[0114] Where M ij represents the matching degree between the i-th robot and the j-th task, σ(·) represents the Sigmoid function, which normalizes the output to the interval [0,1], α and β are matching weights, and R i represents the state vector of the i-th robot, T j represents the task requirement vector of the jth task, W ij It represents the path length from the i-th robot to the location of the j-th task area, and P represents the complexity of the fin structure.

[0115] It should be noted that α and β are obtained through training the support vector machine. During the training process, the success rate of task execution, execution time, resource consumption, etc. are used as evaluation indicators, and the values of α and β are continuously adjusted so that the matching function can more accurately reflect the actual matching situation between the robot and the task.

[0116] Based on the calculated matching degree, the matching status of each robot and task is sorted. The higher the matching degree, the more suitable the robot is for performing the task. For example, for a group of tasks and robots, after calculating all the matching degree values, the robots corresponding to each task are arranged from high to low according to the matching degree to form a matching degree list. By comprehensively considering the ability matching degree and the motion path cost, and using the trained and optimized matching degree function, the matching degree of each robot and task can be accurately quantified. The sorted matching degree list provides a clear reference for task allocation, so that tasks can be more reasonably allocated to suitable robots, improving the efficiency of task execution and resource utilization.

[0117] Specifically, if Figure 4 As shown, the coordination logic of the robot's working sequence includes:

[0118] Arrange tasks in descending order of urgency to generate a time window sequence;

[0119] Determine the temporal continuity of each robot as it completes the time window sequence;

[0120] The tasks that each robot completes in the time window sequence are determined based on the matching degree and time continuity between each robot and the task, so as to coordinate the working order of the robots.

[0121] In order to reasonably arrange the robot's working time and task execution order, the tasks are sorted according to the task urgency and divided into time window sequences, so that tasks can be assigned in different time windows according to the matching degree between the robot status and the task, thereby improving the orderliness and efficiency of task execution; all tasks are sorted in descending order of task urgency. For example, for a set containing multiple fin cleaning tasks, the task urgency of each task is first calculated, and then sorted according to the size of the task urgency. According to the number of tasks and the estimated total execution time, the sorted tasks are divided into several time window sequences {w1,w2,...w, n},w n represents the nth time window, where the length Δt of each time window can be adjusted according to the actual situation, such as being set to a fixed time length, or dynamically adjusted according to the task urgency and the expected execution time. For example, for tasks with high urgency, a shorter time window can be assigned to ensure that they are completed as soon as possible; for tasks with low urgency, a longer time window can be assigned. The generated time window sequence divides the tasks in an orderly manner according to their urgency, making task execution more planned. Reasonable time window settings can ensure that urgent tasks are processed first, while also making full use of the robot's resources and improving the overall task execution efficiency.

[0122] In multi-robot collaborative operations, considering the temporal continuity of robots when completing tasks within a time window sequence can avoid robots from frequently switching tasks, reduce the time and resource consumption caused by task switching, and improve the work efficiency and stability of robots. For each robot, its task allocation in the time window sequence is analyzed. Assume that robot i is in time window w. k and w k+1 There are tasks in both, calculate the time interval Δt between the two tasks k,k+1 , if the time interval Δt k,k+1 If the time interval is less than the set time threshold, it is considered that the task of robot i between the two time windows has good time continuity; otherwise, it is considered that the time continuity is poor. At the same time, the preparation time and movement time required for the robot to complete the task are taken into account. For example, when the robot moves from one task area to another task area, its movement time needs to be calculated and included in the judgment of time continuity. If the movement time is too long, resulting in a time interval greater than or equal to the time threshold, it is considered that the time continuity is affected. Therefore, by judging time continuity, the time utilization efficiency of the robot in the process of executing the task can be evaluated, and the existing task switching problems can be discovered. For cases with poor time continuity, the task allocation can be adjusted in time to reduce unnecessary task switching and improve the work efficiency and stability of the robot.

[0123] Based on the matching degree between each robot and the task and the time continuity, the task completed by each robot in the time window sequence is finally determined, and an efficient collaborative working order between multiple robots is achieved to ensure that the tasks can be completed on time and with high quality; a task allocation matrix is established, with rows representing robots and columns representing tasks. For each time window, based on the matching degree between the robot and the task and the time continuity, a suitable robot is selected from the task allocation matrix to perform the task, with priority given to robots with high matching degree and good time continuity. If there are multiple robots with good matching degree and time continuity, further screening can be performed based on other factors (such as the remaining power of the robot and the current load, etc.). Once the task of each robot in each time window is determined, a corresponding task execution plan is generated, clarifying the task start time, end time and execution order of each robot, and the task execution plan is sent to each robot, and the robot performs the task according to the plan.

[0124] During the task execution process, the progress of the task and the status of the robot are monitored in real time. If a robot is found to have a fault or the task execution progress is delayed, the task allocation and work sequence are adjusted in time, such as reallocating the delayed task to other available robots, or adjusting the task priority to ensure that the overall task can be completed on time; thereby, by comprehensively considering the matching degree and time continuity to determine the task and coordinate the work sequence, efficient collaborative operation between multiple robots is achieved. Reasonable task allocation and work sequence arrangement can give full play to the advantages of each robot, improve the efficiency and quality of task execution, and also enhance the reliability and robustness of the system. Effective work sequence coordination ensures that the fin cleaning material conveying system can operate efficiently and stably. Reasonable task allocation and robot work sequence arrangement can improve cleaning efficiency and reduce resource waste.

[0125] The path control module is used to extract the task instructions of each robot, determine the task position of each robot, and dynamically plan the path and speed of each robot based on the obstacle position of the contaminated area map to control each robot to perform the task.

[0126] Specifically, the planning logic for each robot's path and speed includes:

[0127] Extract the task instructions of each robot to determine the task position and starting position of each robot, and combine the boundary information of the working area in the defaced area map to generate the initial path and initial speed of each robot;

[0128] Adjust the dwell time and coverage of the initial path according to the difficulty of fin cleaning;

[0129] The initial path and initial speed are optimized according to the complexity of the fin structure to obtain the path and speed of each robot;

[0130] The position information of each robot is obtained in real time, and the path and speed of each robot are shared. It is determined whether there is a path conflict to update the path and speed of each robot.

[0131] Before the robot performs a task, it needs to determine a preliminary action plan based on the task instructions and environmental information, that is, generate an initial path and initial speed, provide a basic framework for subsequent path optimization, and ensure that the robot can move towards the task goal; parse the task position and starting position from the robot's task instructions, such as the task instructions contain these position information in a specific data format, extract the relevant coordinate values by writing a parsing program, combine the boundary information of the working area in the contaminated area map, and use the A* algorithm to generate an initial path from the task position to the starting position. In the implementation process of the A* algorithm, the boundary of the working area is set as a passable boundary, and the obstacle position is set as an inaccessible node, guiding A* to move forward. *The algorithm searches for a path that avoids obstacles, and the initial speed is set based on the type of robot and the urgency of the task. For urgent tasks, the initial speed is appropriately increased. For routine tasks, it is adjusted according to the robot's rated operating speed and the complexity of the work area. For example, in a relatively open area with urgent tasks, the initial speed is set to 70% of the robot's maximum speed, while in a narrow and complex work area, the initial speed is set to 40% of the maximum speed. This quickly generates a preliminary action path and speed plan for the robot from the starting position to the task position, enabling the robot to start executing the task in an orderly manner, providing a basis for subsequent path optimization, and improving the system's response speed.

[0132] The fin cleaning difficulty in different areas is different. In order to ensure the cleaning effect, the initial path needs to be adjusted according to the cleaning difficulty so that the robot can perform cleaning work more fully in the working area with a complex environment; the fin cleaning difficulty of the task area is obtained from the contaminated area map. For areas with higher cleaning difficulty, additional stop points are inserted on the initial path. For example, a path interpolation algorithm is used to insert a stop point at a certain distance (such as 0.5 meters) in the high-difficulty area to increase the robot's stay time in the high-difficulty area. At the same time, the robot's travel speed in these high-difficulty areas is reduced, such as reducing the speed to 30% of the initial speed to ensure that the robot has enough time and energy to perform deep cleaning in the area; In areas with lower cleaning difficulty, the stop points on the initial path are appropriately reduced or the robot's travel speed is increased, such as deleting some unnecessary intermediate points to make the initial path simpler, or increasing the speed to 90% of the initial speed to improve cleaning efficiency; By adjusting the path according to the cleaning difficulty, the robot can adopt different cleaning strategies in task areas of different difficulty levels, improving the cleaning effect and efficiency, ensuring that high-difficulty areas are fully cleaned and low-difficulty areas are quickly cleaned.

[0133] The structural complexity of the fin will affect the movement and cleaning operation of the robot. In order to enable the robot to work more efficiently and safely in fin areas with different structures, the path and speed need to be optimized according to the structural complexity; the fin structural complexity is extracted from the contaminated area map, such as the corrugation period of the corrugated fin and the window density of the windowed fin. For the corrugated fin, the curvature of the initial path is adjusted according to the corrugation period. When the corrugation period is small (obtained by comparing with the set period threshold), the curvature of the initial path is reduced accordingly to adapt to the tight corrugated structure of the fin and avoid the robot colliding with the fin. At the same time, according to the influence of the corrugated structure on the cleaning effect, the speed of the robot is adjusted. In the tight corrugated area, the speed is reduced to 50% of the initial speed and the speed is reduced to 50%. The movement frequency of the robot's cleaning tool matches the corrugation period to improve the cleaning effect; for window-shaped fins, the path is optimized according to the window density. In areas with high window density (obtained by comparing with the set density threshold), the path planning is more flexible to make full use of the window space for cleaning, and at the same time, the speed is appropriately increased, such as increasing it to 80% of the initial speed to speed up the cleaning process. In addition, according to the shape and size of the window, the angle and position of the robot's cleaning tool are adjusted to ensure that the inside of the window can be effectively cleaned; thereby, by optimizing the path and speed according to the complexity of the fin structure, the robot's work efficiency and cleaning quality in fin areas with different structures are improved, the risk of collision between the robot and the fin is reduced, and the service life of the robot and cleaning tool is extended.

[0134] In a multi-robot collaborative operation environment, the paths of robots may conflict. In order to avoid collisions and ensure the smooth execution of tasks, it is necessary to detect path conflicts in real time and update the paths and speeds in a timely manner. Each robot shares its position information and corresponding paths and speeds in real time through a wireless communication module. A conflict detection algorithm based on time windows and spatial distance is used to assign a time window [t start ,t end ], where t start Indicates the start time of the robot's task execution, t end It indicates the end time of the robot's task execution and sets a spatial distance threshold. When the spatial distance between the path points of two robots within the time window is less than the spatial distance threshold, it is determined that there is a path conflict.

[0135] Once a path conflict is detected, a priority-based conflict resolution strategy is adopted. The priority is determined by comprehensively considering factors such as the robot's task urgency, remaining power, and distance to the task location. For example, a robot with high task urgency and sufficient remaining power has a higher priority. For low-priority robots, a local path replanning algorithm is adopted, such as re-using the A* algorithm to search for a new path near the conflict point to avoid the conflict area, and adjust the speed according to the new path. This effectively avoids path conflicts among multiple robots during task execution, improves the safety and efficiency of multi-robot collaborative operations, and ensures the smooth progress of cleaning tasks. Real-time path conflict detection and path replanning ensure that the path control module can dynamically adjust the robot's path and speed according to the actual situation of multi-robot operations, so that the entire fin cleaning material conveying system can operate stably and efficiently, improving the reliability and practicality of the system.

[0136] The feedback monitoring module is used to monitor the working status of each robot in performing tasks. When an abnormal working status occurs, it provides feedback and updates the contaminated area map and the state vector of each robot.

[0137] Specifically, the update logic of the defaced area map includes:

[0138] Monitor whether there are new obstacles to update the obstacle location;

[0139] After each robot performs a task, the contaminant thickness and fin surface pressure in the task area are monitored;

[0140] According to the change of the thickness of the pollutants in the task area, it is judged whether the task area is an abnormal working area, and the degree of fin contamination is adjusted in combination with the change of the fin surface pressure in the task area;

[0141] If the task area is an abnormal working area twice in a row, the difficulty of fin cleaning will increase.

[0142] During the robot's mission, the environment will change and new obstacles will appear. In order to ensure the safe operation of the robot and the smooth execution of the mission, it is necessary to timely monitor and update the obstacle location information in the contaminated area map; multiple cameras are arranged in the working area to monitor environmental changes in real time. The camera uses image recognition algorithms to analyze the acquired images and identify new obstacles, such as fallen tools and new equipment parts. When a new obstacle is detected, the camera and original lidar data are fused through a data fusion algorithm to accurately determine the position and shape of the obstacle. The information of the new obstacle is then updated to the contaminated area map and marked as an unreachable area. The size of the working area is simultaneously updated, and the relevant path planning and task allocation information is updated; thereby, the new obstacle information is discovered and accurately updated, ensuring the safety of the robot during work, avoiding collision accidents caused by changes in obstacles, and improving the system's adaptability to environmental changes.

[0143] In order to understand the cleaning effect after the robot performs the task, it is necessary to monitor the changes in the contaminant thickness and fin surface pressure in the task area, so as to evaluate the task execution and adjust the cleaning strategy; through the sensor array, the contaminant thickness and fin surface pressure in the task area are obtained in real time during the robot's task execution. The change in contaminant thickness refers to the absolute value of the difference between the contaminant thickness after cleaning and the contaminant thickness before cleaning, and the change in fin surface pressure refers to the absolute value of the difference between the fin surface pressure after cleaning and the fin surface pressure before cleaning; thereby, the changes in contaminant thickness and fin surface pressure in the task area during the cleaning process can be obtained in real time and accurately, providing direct data support for evaluating the cleaning effect and judging the status of the task area.

[0144] Based on the changes in contaminant thickness and fin surface pressure in the task area, the task area's operating status is determined to be normal. If abnormal, the fin contamination level is adjusted promptly to more accurately reflect the actual situation in the task area and provide a basis for subsequent decision-making. A contaminant thickness change threshold and a fin surface pressure change threshold are set. When the contaminant thickness change in the task area exceeds the contaminant thickness change threshold and the fin surface pressure change exceeds the fin surface pressure change threshold, the task area is determined to be abnormal. For abnormal working areas, the fin contamination level is adjusted based on the direction and degree of change in the contaminant thickness and surface pressure. For example, if the contaminant thickness increases and the fin surface pressure increases, it indicates poor cleaning effect, and the fin contamination level should be appropriately increased. If the contaminant thickness decreases but the surface pressure fluctuates abnormally, further analysis is required and the fin contamination level assessment should be adjusted. The fin contamination level is adjusted according to pre-set grading standards, such as from mild contamination to moderate contamination. This allows for timely detection of abnormal working conditions in the task area and reasonable adjustment of the fin contamination level based on actual conditions, so that the contamination area map more accurately reflects the actual status of the task area and provides more reliable information for subsequent task allocation and path planning.

[0145] If the task area is determined to be an abnormal working area twice in a row, it means that the cleaning difficulty of this area has been underestimated, and the fin cleaning difficulty needs to be increased so that the system can adopt more effective cleaning strategies and resource allocation plans. A counter is set. When the task area is determined to be an abnormal working area, the counter is incremented by 1. When the counter reaches 2, the fin cleaning difficulty increase mechanism is triggered. Ways to increase the fin cleaning difficulty include adjusting the parameters in the cleaning difficulty calculation model, such as adjusting the weight in calculating the fin contamination degree, or directly increasing the level of fin cleaning difficulty. The adjusted fin cleaning difficulty is updated to the contaminated area map, and the relevant task allocation and path planning information is simultaneously updated. For example, the task allocation module is notified to re-evaluate the matching degree between the task and the robot in this area, and the path control module optimizes the robot's path and speed according to the new cleaning difficulty. In this way, the fin cleaning difficulty can be dynamically adjusted according to the task execution status, enabling the system to more reasonably cope with complex cleaning tasks, improve cleaning effect and efficiency, and optimize resource allocation. Reasonable fin cleaning difficulty adjustment enables the feedback monitoring module to work closely with other modules to optimize the operation of the entire fin cleaning material conveying system according to actual working conditions, thereby improving the system's intelligence level and adaptability.

[0146] Example 2:

[0147] like Figure 5 As shown, a method flow chart of a method for allocating fin cleaning material delivery tasks is provided for an embodiment of the present application, the method comprising:

[0148] The sensor array acquires the area location, contaminant thickness, and fin surface pressure, which are then integrated to construct a contamination area map. The contamination area map includes the location of the obstacle, the size of the working area, the degree of fin contamination, and the difficulty of fin cleaning.

[0149] Obtain the remaining power, joint availability, and area reachability of each robot in real time to construct the state vector of each robot;

[0150] Divide the defaced area map into tasks according to the area location, determine the matching degree between each robot and the task, generate task instructions for the robot based on the matching degree, and coordinate the working order of the robots;

[0151] Extract the task instructions of each robot, determine the task location of each robot, and dynamically plan the path and speed of each robot based on the obstacle location of the contaminated area map to control each robot to perform the task;

[0152] Monitor the working status of each robot in performing tasks. When an abnormal working status occurs, feedback is provided and the contaminated area map and the state vector of each robot are updated.

[0153] Since the principle of solving the problem by the method in the embodiment of the present application is similar to that of the system described above in the embodiment of the present application, the implementation of the method refers to the implementation of the system, and the repeated parts will not be repeated.

Claims

1. Fin cleaning material conveying system, characterized in that, include: Modal perception module, task allocation module, path control module and feedback monitoring module; The modal perception module is used to obtain the area position, contaminant thickness and fin surface pressure through the sensor array, and fuse the area position, contaminant thickness and fin surface pressure to construct a contamination area map. The contamination area map includes the obstacle position, the size of the working area, the degree of fin contamination and the difficulty of fin cleaning. The contaminant thickness is obtained by visual detection. The task assignment module is used to obtain the remaining power, joint availability, and area accessibility of each robot in real time to construct a state vector for each robot, divide the tasks according to the location of the area in the defaced area map, determine the matching degree between each robot and the task, generate task instructions for the robot based on the matching degree, and coordinate the working order of the robots; The path control module is used to extract the task instructions of each robot, determine the task position of each robot, and dynamically plan the path and speed of each robot based on the obstacle position of the contaminated area map to control each robot to perform the task; The feedback monitoring module is used to monitor the working status of each robot in performing tasks, and when an abnormal working status occurs, feedback is provided to update the contaminated area map and the state vector of each robot.

2. The fin cleaning material delivery system according to claim 1, characterized in that: The construction logic of the contaminated area map includes: The zone location, contaminant thickness, and fin surface pressure obtained by the sensor array are aligned in time and space, and the obstacle location is identified at the zone location to mark the inaccessible area and the working area, and the size of the working area is determined; In the working area, the degree of fin contamination is comprehensively judged based on the thickness of the contaminants and the fin surface pressure, and the complexity of the fin structure in the working area is extracted; The difficulty of fin cleaning is determined by the degree of fin contamination and the complexity of the fin structure.

3. The fin cleaning material delivery system according to claim 2, characterized in that: The construction logic of the state vector of each robot includes: By monitoring the power of each robot, the remaining power of each robot can be obtained; Monitor the joint angle deviation of each robot to obtain the joint availability of each robot; The remaining power, joint availability, and area reachability of each robot are weightedly fused to obtain the state vector of each robot.

4. The fin cleaning material delivery system according to claim 3, characterized in that: The logic for obtaining the area reachability of each robot includes: Combined with the obstacle locations in the contaminated area map, a safe path is searched for each robot within the working area. Divide the safe path into multiple points and calculate the minimum distance from each point on the safe path to the obstacle location; The safety threshold of each robot is set according to its structure, and the accessibility coefficient of each robot in the working area is determined according to the minimum distance from each point to the obstacle position and the safety threshold of each robot; The fitness factor of each robot is determined according to its structure, and the regional reachability of each robot is obtained by comprehensive judgment based on the reachability coefficient of each robot in the working area and the fitness factor of each robot.

5. The fin cleaning material delivery system according to claim 4, characterized in that: The logic for determining the matching degree between each robot and the task includes: Divide the contaminated area map into tasks according to the area location and construct a task requirement vector, which includes the task cleaning difficulty and task urgency; Determine the capability matching based on the task requirement vector and the state vector of each robot; Determine the path length of each robot to the task area, and combine the fin structure complexity to obtain the motion path cost of each robot; The matching degree of each robot to the task is judged based on the matching degree of capabilities and the motion path cost of each robot.

6. The fin cleaning material delivery system according to claim 5, characterized in that: The coordination logic of the robot's working sequence includes: Arrange tasks in descending order of urgency to generate a time window sequence; Determine the temporal continuity of each robot as it completes the time window sequence; The tasks that each robot completes in the time window sequence are determined based on the matching degree and time continuity between each robot and the task, so as to coordinate the working order of the robots.

7. The fin cleaning material delivery system according to claim 6, characterized in that: The planning logic for each robot's path and speed includes: Extract the task instructions of each robot to determine the task position and starting position of each robot, and combine the boundary information of the working area in the defaced area map to generate the initial path and initial speed of each robot; Adjust the dwell time and coverage of the initial path according to the difficulty of fin cleaning; The initial path and initial speed are optimized according to the complexity of the fin structure to obtain the path and speed of each robot; The position information of each robot is obtained in real time, and the path and speed of each robot are shared. It is determined whether there is a path conflict to update the path and speed of each robot.

8. The fin cleaning material delivery system according to claim 7, characterized in that: The update logic of the defaced area map includes: Monitor whether there are new obstacles to update the obstacle location; After each robot performs a task, the contaminant thickness and fin surface pressure in the task area are monitored; According to the change of the thickness of the pollutants in the task area, it is judged whether the task area is an abnormal working area, and the degree of fin contamination is adjusted in combination with the change of the fin surface pressure in the task area; If the task area is an abnormal working area twice in a row, the difficulty of fin cleaning will increase.

9. The fin cleaning material delivery system according to claim 8, characterized in that: The distance from each point on the safe path to all obstacle locations in the contaminated area map is calculated using the Euclidean distance formula, and the minimum value of the distance is selected as the minimum distance from each point to the obstacle location.

10. The fin cleaning material delivery system according to claim 9, characterized in that: The sub-logic for obtaining the joint availability of each robot includes: Monitor the actual angle of each joint in real time, calculate the absolute value of the difference between the actual angle of each joint and the preset nominal angle, and obtain the joint angle deviation of each joint; Calculate the ratio of the joint angle deviation of each joint to the maximum angle deviation of each joint to obtain the joint deviation of each joint; Subtract the joint deviation of each joint from 1 to obtain the joint availability of each joint; The joint availability of all joints of each robot is averaged to obtain the joint availability of each robot.

11. Fin cleaning material delivery task allocation method, characterized in that: include: The sensor array acquires the area location, contaminant thickness, and fin surface pressure, which are then integrated to construct a contamination area map. The contamination area map includes the location of the obstacle, the size of the working area, the degree of fin contamination, and the difficulty of fin cleaning. Obtain the remaining power, joint availability, and area reachability of each robot in real time to construct the state vector of each robot; Divide the defaced area map into tasks according to the area location, determine the matching degree between each robot and the task, generate task instructions for the robot based on the matching degree, and coordinate the working order of the robots; Extract the task instructions of each robot, determine the task location of each robot, and dynamically plan the path and speed of each robot based on the obstacle location of the contaminated area map to control each robot to perform the task; Monitor the working status of each robot in performing tasks. When an abnormal working status occurs, feedback is provided and the contaminated area map and the state vector of each robot are updated.

Images